On Wed, 2024-02-21 at 09:09 +0100, Johannes Berg wrote: > > I guess that'd also apply to probe responses? Or does it not send > those > at all? But we discussed before that maybe we don't have to be > perfect > there, so I guess we can find some solution to that. > > So we have, so far > > Realtek: > - uses offset in beacon template to set partner counter, > - requires host to set CSA/ECSA elements > > Qualcomm: > - copies and updates CSA/ECSA elements all by itself > - btw, not sure here about probe responses, does it do that too? > > hwsim: > - personally, I'd prefer it works like Realtek, for less complexity > > Intel: > - not implemented, probably not going to happen any time soon, but I > think we might prefer the Realtek way too if we ever do this > > > Mediatek? Broadcom? Anyone else? > > > I guess then for Qualcomm we'll need to just add an extended feature > flag for "FW_HANDLES_PARTNER_LINK_CSA" and then hostapd can just not > update the parter link templates, or so? > > Thanks, Hi, Sorry for the late reply. In summary, MediaTek firmware behaves similarly to Realtek's firmware, so we agree on the conclusion of this thread. By the way, this behavior does not apply to probe response in MediaTek firmware. In other words, MediaTek firmware does not modify the probe response and it requires hostapd to fill all CSA-related Elements. I think this could be fixed if hostapd provides CSA offsets when sending the probe response. Best, Michael > johannes
